GLACIATION OF PRINCE WILLIAM SOUND 479 Passage, where glacial erosion would have been most effective. The low rock benches on both sides of Port Nellie Juan near the mouth are not completely flat-topped, but have gently-undulating surfaces, as if modified by glaciation. The lack of seacliffs and talus slopes at the backs of these terraces suggests that they are of preglacial origin and not uplifted, wave-cut, post-glacial benches. There are two groups of flat-topped islets, the first between Montague and Green Islands, the second, called Porpoise Rocks, in Port Etches at the west end of Hinchinbrook Island. These are in places where remnants of the rock floor of Prince William Sound might well be preserved, because they are out of the paths of the most rapidly moving ice currents of the former piedmont glacier of Prince William Sound. The borders of each islet are cut into precipitous cliffs, and, as they are located where wave-work adequate for such erosion is now impossible, these cliffs are ascribed to waves from the iceberg-discharging front of the retreating Prince William Sound Glacier. Recent Changes of Level in Prince William Sound. Mention has already been made of the post-glacial uplift on the continental shelf near Prince William Sound, and also a short distance to the east on Wingham Island, and of the shore forms near Columbia Glacier which Grant and Higginsx ascribe to recent uplift, but which we believe to be due to the work of iceberg waves. The latest change of level of the land adjacent to Prince William Sound, however, is a very slight downward movement. The evidence that the latest change of level was a sinking of the land has been brought forward by Vancouver,8 Hayes,8 Schrader4 and by Grant and Higgins.5 Our own additional observations, made in 1910, are that there are dead trees in place and evidently killed by post-glacial submergence in three other localities. These are (1) on Point Countess, Flemrm'ng Island, on the southwestern side of lower Knight Island Passage, (2) on the southeastern side of Green Island, and (8) at Graveyard Point on the northwestern side of Montague Island. At each of these places the submergence may be still going on. These localities are 6 to 32 miles apart and are 50 and 80 miles west of the areas of recent sinking on the Copper River delta and at Bering Lake, showing that the downward movement is widespread. On the other hand, it is extremely localized, for the areas of dead trees are very small and there is undisturbed, continuous forest close to sea level between these points of submergence, proving that there is no continuous sinking of the land. There is no reason for suspecting that this recent submergence in Prince William Sound is related to glaciation, as it probably is on the Copper River delta where isostasy and the loading of the seacoast with the outwash deposits of the Copper River glaciers may possibly explain the submergence. Glacial Deposits. Most of the glacial deposits in Prince William Sound are submarine. In discussing the several fiord regions we have already described small areas of ground moraine, terminal and lateral moraine, outwash gravels, and deltas on the land, as well as the moraine bars and other deposits which mark terminal moraines in the sea. There > Bull. 443, U. S. Geol. Survey, 1910, p. 18. »See footnote on p. 278 in H. H. Bancroft's History of Alaska, San Francisco, 1886. • Hayes, C. W., Nat Geog. Mag., Vol. IV, 1892, p. 136. • Schrader, P. C., 20th Ann. Kept., TJ. S, Geol. Survey, Part VH, 1900, p. 404. « Grant, U. S. and Higgins, D. F., Bull. 443, TJ. S. Geol. Survey, 1910, pp. 17-18.